How can I characterize locations in the NY/NJ area by urban/suburban?
October 31, 2013 10:09 AM   Subscribe

I have about 400 address in the NY/NJ metro area, and I'd like to characterize each address as urban or suburban based on some standard definition. What is a good way to do this? I'm willing to use a paid service to code my addresses.

I've tried looking at Metropolitan and Micropolitan Statistical Areas and they are not quite granualr enough. By some definitions, including MSA, *all* locations within a big radius of NYC will be considered urban, but I'd like to try to separate these out somehow.
posted by scottso17 to Work & Money (11 answers total) 1 user marked this as a favorite
How about looking up their ZIP code's population using this dataset?
posted by julthumbscrew at 10:29 AM on October 31, 2013

If you have the zip codes or borough names, it should be straightforward- if not in one of the five boroughs of NYC (Manhattan, Queens, Brooklyn, Bronx, Staten Island), it would be a suburban area. You can find which zip codes correspond to the boroughs.

Perhaps I am missing something, though- if you don't have the zipcodes or borough names, not sure how you'd do it then.
posted by bearette at 10:31 AM on October 31, 2013

bearette: "if not in one of the five boroughs of NYC (Manhattan, Queens, Brooklyn, Bronx, Staten Island), it would be a suburban area."

You'd probably want to include Jersey City and Newark in the "urban" areas, as well.
posted by Grither at 10:35 AM on October 31, 2013 [2 favorites]

Somebody who works with GIS could probably do this based on population density, or number of residents per square mile. You'd still have to define what's urban and what's not.
posted by mareli at 10:55 AM on October 31, 2013 [1 favorite]

You're first going to have to define urban and suburban in order to do this. As Grither says, Newark is definitely an urban area that is next to NYC, as is Jersey City. Although many places in this giant metro area are suburban in the sense that people live there but commute to work in NYC, they still feel somewhat urban in the Jersey CIty sense, ie Hoboken and Secaucus. I think you'll have to decide, for example, if your dividing line has to do with population density, amount of private green space, amount of industrial vs residential area, etc. In NJ there are towns called "cities" that are very suburban that are about 20 - 30 miles from NYC -- little cities filled with single family homes and yards that are to some extent NYC bedroom communities despite having their own downtown -- and near them, there are smallish urban-looking "townships" or "towns" closer to NYC that don't feel very suburban if you're thinking of the 1950s suburban ideal. I think you should first decide if you're talking about population density, private green space, residential vs industrial space, etc.
posted by third rail at 11:03 AM on October 31, 2013 [1 favorite]

Yeah. First you have to define urban vs. suburban. Let's say you're using population density.

A) Pay someone to use GIS. This would be a pretty quick thing to do -- they'd geocode the addresses into lat/long, find the relevant census tracts with a spatial join, grab their population density, export as an excel file. That way, you can take a density threshold and filter the addresses as you see fit.

B) You might want to look at Batchgeo, which seems to do geocoding for you. So at least you'd be able to convert your addresses to lat/long coords and do some own editing yourself.

It really depends on the criterion, though. If you're using 'commute distance', I believe that the ACS census has county-resolution data about distance to commute. If you're using 'availability/usage of public transportation' (which is my personal definition of suburban/urban), I recall that information being in some census dataset, also.
posted by suedehead at 11:19 AM on October 31, 2013

Unless you're being very literal or need very clean criteria (NYC vs. all other municipalities in the region), I'd shy away from just putting NYC as urban overall; some parts of outer Queens (say, Douglaston) and much of Staten Island are very suburban in character, while "suburbs" like Hoboken and Jersey City feel much more urban than those places.

Instead, I'd figure out what combination of the following largely quantifiable criteria you use to define cities and suburbs:

- Population density/residents per square mile
- Average lot size (for houses?) or proportion of single-family vs. multi-family housing
- Parking requirements, percentage of land devoted to parking
- Presence or percentage of mixed-use developments
- Mode-share: use of public transportation, single-occupancy car, etc., or geographic access to public transportation lines
posted by andrewesque at 12:30 PM on October 31, 2013

If you have access to GIS software, the census bureau has (free) shapefiles prejoined with demographic data. At a glance, some of your criteria might include:

Means Of Transportation To Work (more car usage = more likely suburban)
Travel Time To Work (longer commute = more likely suburban)
Housing Units (lower number of units in the designated area = more likely suburban)
Units In Structure (lower number of units per structure = more likely suburban)
Median Year Structure Built (this can be really confounding if you're not familiar with the area, but in general, newer = more likely suburban)
Bedrooms (higher number = more likely suburban, but I don't know the NYC area well enough to be confident on this)

BatchGeo will just make a map with markers, which is not what you need. You need a list that spits out lat/long so you can join it to your TIGER file. Here is a list of geocoding services (free and paid). The trial version of this one worked fine for me.
posted by desjardins at 12:57 PM on October 31, 2013

Here is a FAQ relating to how the Census defines "urban areas."
posted by desjardins at 12:59 PM on October 31, 2013

In NY, Mount Vernon, Yonkers and New Rochelle would all be urban.
posted by mlis at 1:10 PM on October 31, 2013

Might be overkill but UrbanMapping's API could likely do this sort of thing. Used it at a previous job.
posted by evisceratordeath at 4:18 PM on October 31, 2013

« Older Getting my Numbers to Stand up Straight   |   Why labor induction if everything is going well? Newer »
This thread is closed to new comments.